AMD GPU - How to use other model (e.g. NovelAI, CivitAI ckpt ) Stable Diffusions, method 2

this video gives a demo of how to use other model ckpt files for AMD GPU running stable diffusion on Windows.
👉ⓢⓤⓑⓢⓒⓡⓘⓑⓔ Thank you for watching! please consider subscribe. thank you!
(I had a dental visit today and swelling teeth, so not able to speak well. But want to share the tutorial with everyone ASAP. Thus the later part of video only have some music background. Sorry about it.)
1:24 fetch the ckpt model file.
1:53 download the needed script. Links:
2:45 activate python environment
3:30 Convert stable diffusion (.ckpt) to diffusers
4:38 Convert diffusers model to ONNX format model
5:54 Test run of the converted model
7:44 Other models (such as ckpt files from Civitai that you can try.
If you need troubleshooting, please post them in comments. Good luck!

[TheEveryoneGames]

this is a relly good tutorial, thanks for the easy to follow instructions. gonna put my 6700xt to good use.

[mikhailpodgornyak]

Thank yoy very much!! Protogen is also the one i really want to play with

[freitasLeafar]

Very good tutorial, but for AMD cards I would advise to do a dual boot, the difference in performance is worth it

[ZiosNeon]

hey man very awesome, can you take a look at safetensors? it is getting popular it would be awesome to be able to convert them as well.

[Smarglenargle]

Now show us how to disable the censor 😈

[aprivatechannel]

But where did you get the v1 and v2 interface.yaml? Is it possible if the file only comes with a .ckpt? I still get errors when i try to just convert a .ckpt file

[fhuckgoogle]

When I type conda activate sd39 it says "EnvironmentNameNotFound: Could not find conda environment: sd39"

I was able to install using your previous video but it seems now I'm stuck on the first step.

[moshegoldbergsteinrabbinow]

I get "ModuleNotFoundError: No module named 'diffusers'" when i try to convert to diffusers

[mikhailpodgornyak]

Seems like this script doesn't work with safetensor((
Do I understand it correctly, that ckpt is pickletensor?
Do you think it is possible to upgrade your script to work directly with safetensor (safetensor -> diffusers), or should I look for a way to convert to pickle first?

[jzamoral]

Hello, do you have the How to start video? Im so noob but im interesting to run at my pc

[Pixeljunk]

Do you know of any way to use the AUTOMATIC1111 webui script with amd cards? My knowledge and understanding of this topic isnt good enough for me to get it to work.

[aprivatechannel]

Hey! Im a bit new, but if I have a model that comes with a .vae file, what should i do?

[fartboobs]

When I run the 'convert_stable_diffusion_to_checkpoint_to_onnx.py', I get the error: "Exporting the operator to ONNX opset version 14 is not supported."

Any ideas?

[TerenceThyr]

This happens to come up for me when converting from ckpt to diffusers. Can I add this into the Python Script? If so, where? Sounds like it would be better for the model but with this all being said, should I even be concerned about it?

Checkpoint has both EMA and non-EMA weights.
In this conversion only the non-EMA weights are extracted. If you want to instead extract the EMA weights (usually better for inference), please make sure to add the `--extract_ema` flag.

[Cewelowic]

Is there a way to use the AMD Web UI to have more than 77 tokens? Trying to recreate other peoples prompts seems to truncate the input

[ZannaBianca]

Hello!!! I am following your great tutorial, but I have a problem I don't know how to fix:

After the command Convert stable diffusion (.ckpt) to diffusers I have this error

RuntimeError: [enforce fail at data. DefaultCPUAllocator: not enough memory: you tried to allocate 16777216 bytes.

And the conversion what can I do to fix it?

Thank you very much!!!

[nodewizard]

Does this method work for ".safetensors" ?
How can you use the CivitAI .safetensors files. Thank you

[mikhailpodgornyak]

I have a problem - most of the converted models do not produce any results - random pattern. Do you know how could it be fixed?